Reinhard Tartler <[email protected]> writes: > On Wed, Aug 24, 2011 at 20:13:43 (CEST), Måns Rullgård wrote: > >> Reinhard Tartler <[email protected]> writes: >> >>> This avoids a segmentation fault in hyscale_fast_MMX2 as it tries to >>> call code memory that is not marked as executable. >>> >>> Reported and Analyzed by: Colin Watson <[email protected]> >>> --- >>> configure | 4 ++-- >>> 1 files changed, 2 insertions(+), 2 deletions(-) >>> >>> diff --git a/configure b/configure >>> index d4e352e..5baf327 100755 >>> --- a/configure >>> +++ b/configure >>> @@ -2466,7 +2466,7 @@ case $target_os in >>> add_cppflags -U__STRICT_ANSI__ >>> ;; >>> linux) >>> - add_cppflags -D_POSIX_C_SOURCE=200112 -D_XOPEN_SOURCE=600 >>> + add_cppflags -D_POSIX_C_SOURCE=200112 -D_XOPEN_SOURCE=600 >>> -D_SVID_SOURCE >> >> Does this have any potentially unpleasant side-effects? > > It does have the pleasent side-effect of avoiding a segmentation fault > in swscale, and still passes a fate run. On this basis, I'm therefore > inclined to say 'no'.
That is not an answer. The question is what else it enables (or disables) in glibc. These things are not to be toyed with. -- Måns Rullgård [email protected] _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
